Multi-Linear Kernel Regression and Imputation in Data Manifolds

YNAMIC magnetic resonance imaging (dMRI) is a popular non-invasive imaging modality for observing regularizers which are widely used in manifoldlearning body organ movement, with rich potential in cardiac and approaches [16]-[18]. MultiL-KRIM adopts instead a neurological diagnosis [1]. DMRI stands out as an application "collaborative-filtering" modeling approach to identify "optimal" domain where regression grapples with all of the and manifold-cognizant combinations of the observed archetypal data-analytic bottlenecks: large dimensionality due data features for regression and imputation. MultiL-KRIM to the image data, dynamic data patterns due to dMRI's time needs no training data to operate, builds a nonparametric component, missing data due to under-sampling, and strong regression estimate to reduce the dependence of its modeling but unknown spatio-temporal correlations since, often, dMRI assumptions on the probability distribution of the data [37], monitors structured movement; e.g., a beating heart [2].